PCA

目前MATLAB里边对于多点计算仿射矩阵普遍采用RANSAC法，但是比较麻烦计算量也相对较大，通过最小二乘法来进行仿射矩阵的估计能够大大提高该效率，所以奉上自己在实验室写的仿射矩阵估计函数，来弥补matlab函数库在这方面的不足。(The MATLAB inside for calculating affine matrix commonly used RANSAC method, but more trouble calculation is relatively large, estimated by the least squares method to the affine matrix can greatly improve the efficiency, so give yourself in laboratory to write the estimated affine matrix function, to compensate for the lack of MATLAB Library in this area.)

立体视觉代码。主要过程包括：harris角点检测，zncc最优策略匹配，扩散得到更多特征对应，计算disparity和对应颜色值。另外提供了两幅图片以及对应的特征匹配的结果和disparity图像。(Three-dimensional visual codes. The main process includes: harris corner detection, zncc optimal strategy for matching, the proliferation of more features of the corresponding calculation of disparity and the corresponding color value. Also provides a picture and two corresponding feature matching results and disparity images.)

首先通过建立一个可靠的背景更新模型，然后在视频序列中取一帧图像与背景做差得到完整可靠的运动目标图像(This program is used to detect the moving targets by image substraction ,)

给予matlab的JPEG压缩程序，包含原始压缩图片，程序对灰度bmp图片进行JEPG编码，压缩成以0101形式的编码。JPEG编码先把图象色彩RBG变成亮度Y和色度Cr、Cb，它利用人的视觉对色度不敏感的特点，减少一部分色度数据，以达到压缩。 JPEG采取多种编码方式，包含有行程编码(Run Length Coding)和哈夫曼(Huffman)编码，有很高的压缩比。在编码前，先对数据进行分块，离散余弦变换（DCT）及量化，保留能量大的低频信号，丢弃高频信号以达到压缩。(Given to the JPEG compression matlab program, contains the original compressed image, the program of the grayscale bmp picture JEPG encoded, compressed into a 0101 form of encoding. RBG color coded JPEG image into a first luminance Y and chrominance Cr, Cb, which uses the human visual characteristics is not sensitive to the color, reduce the number of color data to achieve compression. Take a variety of JPEG encoding, includes length coding (Run Length Coding) and Huffman (Huffman) coding, a high compression ratio. In the code before the data block, discrete cosine transform (DCT) and quantization, to retain a large low-frequency signal energy, discarding high-frequency signals in order to achieve compression.)

利用脉冲耦合神经网络（pcnn）实现图像滤波的源程序，已经过测试。(Using pulse coupled neural networks (pcnn) the realization of the source image filtering, has been tested.)

中科院图像理解与计算机视觉课件6 值得参考一下 some topics on pattern recognition and computer vision(Chinese Academy of Sciences and Computer Vision Image Understanding Courseware 6 worth refer to some topics on pattern recognition and computer vision)

天津工业大学课程设计，内容包含vs2010对图像的灰度化，二值化，边缘检测。包含单纯c语言编写的对图像的灰度化，二值化，边缘检测的一个.C文件，可以直接用vc6.0运行，也可以直接在DSP上运行，还包含一份详细的设计报告。(Tianjin Polytechnic University curriculum design, content includes vs2010 grayscale image binarization, edge detection. Contains pure c language for image graying, binarization, edge detection is one. C file that can be run with vc6.0 can also be run directly on a DSP, also contains a detailed design report.)

各向异性扩散相干斑降噪，可以在增强边缘的同时滤出噪声，效果很好！(Anisotropic diffusion, coherent noise can enhance the edge of the filter out the noise at the same time, the effect is very good!)

说明：  利用BP算法进行SAR成像，BP算法是一种对时域回波数据进行成像的算法(BP algorithm was used for SAR imaging)

说明：  Hu的七个不变矩在图像匹配检索中适用广泛。本代码仅计算图像间Hu的七个矩的D2大小，直观的了解Hu矩的应用(seven moment invariants in image retrieval in the application of broad match. Only the calculation of the code between the image of the seven Hu moments D2 size, intuitive understanding of the application of Hu Moment)